SM_plot_roi.m

function roi_ave= SM_plot_roi(ROIS,varargin)
%fb_select_roi selects an arbitrary number of roi's for plotting
%
%
% *** for STEFFEN WOLFE
%
%

colors=eval(['winter(' num2str(length(ROIS.coordinates)) ')']);
sono_colormap='hot';
baseline=3;
ave_fs=30*20;
save_dir='roi';
template=[];
fs=48000;
per=2;
max_row=5;
min_f=0;
max_f=9e3;
lims=1;
dff_scale=20;
t_scale=.5;
resize=1;
detrend_traces=0;
crop_correct=0;

nparams=length(varargin);

if mod(nparams,2)>0
	error('Parameters must be specified as parameter/value pairs');
end

for i=1:2:nparams
	switch lower(varargin{i})	
		case 'colors'
			colors=varargin{i+1};
		case 'sono_colormap'
			sono_colormap=varargin{i+1};
		case 'baseline'
			baseline=varargin{i+1};
		case 'ave_fs'
			ave_fs=varargin{i+1};
		case 'save_dir'
			save_dir=varargin{i+1};
		case 'template'
			template=varargin{i+1};
		case 'fs'
			fs=varargin{i+1};
		case 'per'
			per=vargin{i+1};
		case 'max_row'
			max_row=varargin{i+1};
		case 'dff_scale'
			dff_scale=varargin{i+1};
		case 't_scale'
			t_scale=varargin{i+1};
		case 'resize'
			resize=varargin{i+1};
		case 'detrend_traces'
			detrend_traces=varargin{i+1};
		case 'crop_correct'
			crop_correct=varargin{i+1};
	end
end

if resize~=1
	disp(['Adjusting ROIs for resizing by factor ' num2str(resize)]);

	for i=1:length(ROIS.coordinates)
		ROIS.coordinates{i}=round(ROIS.coordinates{i}.*resize);
	end
end


mkdir(save_dir);

% ROIS is a cell array of image indices returned by fb_select_roi
%

% first convert ROIS to row and column indices, then average ROI and plot
% the time course

% TODO save image with ROIs (using color scheme that's used for time plots)

mov_listing=dir(fullfile(pwd,'*.mat'));
mov_listing={mov_listing(:).name};

to_del=[];
for i=1:length(mov_listing)
	if strcmp(mov_listing{i},'dff_data.mat')
		to_del=i;
	end
end
mov_listing(to_del)=[];

roi_n=length(ROIS.coordinates);

load(fullfile(pwd,mov_listing{1}),'video')%'mic_data','fs');
mov_data = video.frames;
for i = 1:length(mov_data)
     mov_data2(:,:,i) = mov_data(i).cdata;
end
mov_data = double(mov_data2);
[rows,columns,frames]=size(mov_data);

ave_time=0:1/ave_fs:size(mov_data,3)/30;

% need to interpolate the average onto a new time bases

roi_ave.raw={};
roi_ave.interp_dff=zeros(roi_n,length(ave_time),length(mov_listing));
roi_ave.interp_raw=zeros(roi_n,length(ave_time),length(mov_listing));
disp('Generating single trial figures...');
clear mov_data
for i=1:length(mov_listing)

	disp(['Processing file ' num2str(i) ' of ' num2str(length(mov_listing))]);
	load(fullfile(pwd,mov_listing{i}),'video')%,'mic_data','fs','vid_times');
    mov_data = video.frames;
for ii = 1:length(mov_data)
     mov_data2(:,:,ii) = mov_data(ii).cdata(:,:,:,:);
end
mov_data = double(mov_data2);
	% resize if we want

	if resize~=1

		disp(['Resizing movie data by factor of ' num2str(resize)]);

		frameone=imresize(mov_data(:,:,1),resize);
		[new_rows,new_columns]=size(frameone);

		new_mov=zeros(new_rows,new_columns,frames);

		for j=1:frames		
			new_mov(:,:,j)=imresize(mov_data(:,:,j),resize);
		end
		
		%im_resize=im_resize.*resize;
		mov_data=new_mov;
		clear new_mov;

	end

	[path,file,ext]=fileparts(mov_listing{i});
	save_file=[ file '_roi' ];

	% highpass for mic trace

	

	[rows,columns,frames]=size(mov_data);
	roi_t=zeros(roi_n,frames);
% 
% 	if length(frame_idx)~=frames
% 		warning('Trial %i file %s may be corrupted, frame indices %g not equal to n movie frames %g',...
% 			i,mov_listing{i},length(frame_idx),frames);	
% 		frame_idx=frame_idx(1:frames);
% 	end
frame_idx = 0:size(mov_data,3)-1;
	timevec=(frame_idx./30); %movie_fs

	disp('Computing ROI averages...');

	[nblanks formatstring]=fb_progressbar(100);
	fprintf(1,['Progress:  ' blanks(nblanks)]);

	% unfortunately we need to for loop by frames, otherwise
	% we'll eat up too much RAM for large movies

	for j=1:roi_n
		fprintf(1,formatstring,round((j/roi_n)*100));

		for k=1:frames
            
			tmp=mov_data(ROIS.coordinates{j}(:,2),ROIS.coordinates{j}(:,1),k);
			roi_t(j,k)=mean(tmp(:));
		end
	end

	fprintf(1,'\n');

	dff=zeros(size(roi_t));
    
   
	% interpolate ROIs to a common timeframe

	for j=1:roi_n

		tmp=roi_t(j,:);
        
        

		if baseline==0
			norm_fact=mean(tmp,3);
		elseif baseline==1
			norm_fact=median(tmp,3);
		elseif baseline==2
			norm_fact=trimmean(tmp,trim_per,'round',3);
		else
			norm_fact=prctile(tmp,per);
		end

		dff(j,:)=((tmp-norm_fact)./norm_fact).*100;

        
		yy=interp1(timevec,dff(j,:),ave_time,'spline');
		yy2=interp1(timevec,tmp,ave_time,'spline');

		roi_ave.interp_dff(j,:,i)=yy;
		roi_ave.interp_raw(j,:,i)=yy2;

    end

    
    
%save individual files	
%save(fullfile(save_dir,[save_file '.mat']),'roi_t','frame_idx','fs','timevec');



	roi_ave.raw{i}=roi_t; % store for average
	roi_ave.filename{i}=mov_listing{i};
	
end

roi_ave.t=ave_time;
save(fullfile(save_dir,['ave_roi.mat']),'roi_ave');
disp('Generating average ROI figure...');

% plot the averages with confidence intervals

%timevec=ave_time;

% if template is passed use the template mic trace, otherwise use the last song

%roi_mu=mean(roi_ave.interp_dff,3);
%roi_sem=std(roi_ave.interp_dff,[],3)./sqrt(size(roi_ave.interp_dff,3));

%if ~isempty(template)
%	[song_image,f,t]=fb_pretty_sonogram(double(template),fs,'low',1.5,'zeropad',1024,'N',2048,'overlap',2040);	
%end

%fb_multi_fig_save(save_fig,save_dir,'ave_roi','eps,png,fig','res',100);

%%%%%%%%%%%%% CELL MASK MATCHED TO ROI
% plot cell masks color-matched to their ROIs